View on GitHub

IntelliJ code history

This is visualization of IntelliJ community edition project code history. It was made with this IntelliJ plugin and some scripts. There is no particular method behind this so if you have an idea, please feel free to suggest it. (You can find csv file with events and original visualizations on google drive.)

Change size chart

Shows the amount of changes by day/week/month. The idea is to see overall trend of the project. (Similar to github contributors graph but without interpolation.)
Note downward trends around New Year and spikes in November/December which somewhat correlate with major releases.



Amount of committers

Shows how many different people committed over month/week/day. The idea is to see the amount of people contributing to the project.



Amount of files in commit

Shows average amount of files in commit by day/week/month. Assuming that commit is a finished unit of work, the idea is to see how its size changes over time.



Files changed in the same commit

Shows files which were changed in the same commit several times. Thickness of edges is proportional to the number of commits. The idea is to discover de facto dependencies between files.
You can click on nodes to see file names in cluster.



Committers changing same files

Shows commiters connections to files they have both changed within a week. That is if file was modified by two persons within a week, a link is added. The is idea is to discover how people "communicate" through changing same files.



Amount of commits treemap

Shows a break-down of commits by package/folder. Size of rectangles corresponds to amount of commits. The idea is to see which parts of the project have more attention (e.g. ratio between commits to production code and tests).
(Renamed and moved package are tracked but only current package name is displayed.)



Commit time punchcard

Shows amount of commits at certain time of day (similar to github punchcard).
Note that you can also group commits by minute (inspired by subsecond offset heatmaps).



Time between commits histogram

This distribution will probably be the same for any project, i.e. you are most likely to commit again within 5 hours or at least within a day. (Time is tracked separately for each committer.)



Commit messages wordcloud

Some prepositions and articles are excluded as well as some words specific for the project but not related to code. The idea is to see what happened with code by observing human language.